Dispatches from the Crusty Cartographer: Episode One — Square Pegs, Round Holes
Large variations exist between the boundaries used to report census demographics and the study areas to which we’d like to apply them. How, then, do we take the information stored in one geography and apply it to another when their boundaries do not align? Or do we even do it at all?
Build understanding through boundaries
Whether you know it or not, your life operates under an extensive set of boundaries. You move about each day within cities, zip codes, school districts, watersheds, planning areas, and on (and on) it goes. These boundaries are devised to help us digest and manage our world, but to do this, we need to understand what they contain.
For example, school administrators examine the distribution of age within their district so that they can plan for new buildings. Likewise, a mayoral candidate is wise to assess a city’s demographics to help develop their campaign strategy.
When it comes to gathering and reporting the information required to understand a place, it’s often too costly to do this through direct surveys or other on-the-ground techniques. Rather, researchers look to existing sources of information that they can use to make these inferences.
Finding the Missing Piece of Your Puzzle
In the U.S., a common source of data is the Census Bureau, which collects detailed demographics and other statistical data about our nation. At its root, the Census Bureau uses a combination of boundary data called blocks, block groups, and census tracts to aggregate and report the data it collects. While census geographies are valuable, these boundaries do not always align with the areas we’d like to study in our own communities.
Imagine an urban planner who wants to summarize the income statistics reported by the Census Bureau along a transit line for a narrow planning corridor. This seems straight forward at the outset of the work with the question being: What is the average income within the planning corridor of our proposed transit line?
But as soon as she digs into her work, she will likely have difficulty aligning the census boundaries to her planning zone.
It’s a spatial conundrum that leaves our planner feeling as though she’s trying to fit square pegs into round holes.
Summarizing census demographics to custom boundaries is a common challenge when working with spatial data. When confronted with these sorts of issues related to mismatched boundaries there are two broad options at your disposal.
Option 1: Map the data as it is and leave decisions to the visual analysis of the audience.
Option 2: Employ sophisticated forms of spatial analysis that produce quantitative results for your study area.
Base Technique On Purpose
Option one: Often, the simpler map-based approach in which you create a map that overlays your study area on top of the underlying census information will provide an adequate solution for your audience. Through sound map design, the audience can usually glean everything they need to know via the visual cues on the map. With this approach, you’re giving into the data by not transferring information from one set of boundaries to another. Instead, you’re preserving the source data and letting the map tell the story.
Option 2: The more complex approach involves a technique known as spatial apportionment.
In this approach, a set of processes are run against the data to transfer the attributes from the record of source to your area of interest. The essential aspect of the apportionment process is that it captures the areas of overlap between the two sets of boundaries, along with the spatial distribution of key statistics within those areas of overlap. The results of this capture are then used to transfer, or apportion, the statistics from one set of boundaries to the other.
When performed correctly, spatial apportionment can produce reliable estimates within your study area. However, caution is in order when using the technique because it requires a strong background in spatial analysis and statistics along with a thorough understanding of the geographic data being used. Simply put, spatial apportionment is a powerful technique, but not for the faint-hearted.
The Prescription
All said, there is no best approach when it comes to dealing with the misalignment between census geographies and other forms of boundary data — but there’s definitely a ‘better’ approach depending on the circumstances.
Ultimately, the approach chosen should be based on the needs of the audience and the skills of the analyst.
Equally important to all of this is that your audience understands how the information was derived and any potential errors associated with it. If you choose your method wisely, and adequately document the process and its limitations, you’ll derive reliable results useful to your audience — even if it felt like stuffing a square peg into a round hole.
Like this article? Subscribe to our newsletter to read more like it here.
About the Author: Brian Parr has seventeen years of experience working with GIS tools and techniques. He is currently a GIS Project Developer at mySidewalk.
